basic_mcp_server

Basic MCP Example - Weather Service

This is a basic example demonstrating how to implement an MCP (Multi-Cloud Protocol) server using FastMCP. The example implements a simple weather service that showcases the three main MCP components: tools, resources, and prompts.

Purpose

This project serves as a learning example to demonstrate:

• How to create an MCP server using FastMCP
• How to implement MCP tools, resources, and prompts
• Basic MCP server structure and implementation

Implementation Details

The server implements three MCP components to demonstrate different interaction patterns:

1. Tool: get_weather(location: str)

   • Example of a simple MCP tool
   • Shows how to define a tool with parameters
   • Demonstrates tool return values

2. Resource: weather://{location}

   • Example of MCP resource usage
   • Shows resource URL pattern matching
   • Demonstrates resource parameter handling

3. Prompt: weather_report(location: str)

   • Example of MCP prompt implementation
   • Shows how to create formatted prompts
   • Demonstrates prompt parameter usage

Running the Example

To run this example:

1. Install FastMCP library
2. Run server.py directly

Requirements

• Python 3.13 or higher
• MCP library with CLI support (mcp[cli])

Project Structure

• server.py: MCP server implementation
• main.py: Entry point that runs the server
• pyproject.toml: Project configuration
• .gitignore: Git ignore rules

Running the Example

To run this example, follow these steps:

1. Install uv package manager:

   # On macOS/Linux
   curl -fsSL https://uv.rs/install.sh | bash
   
   # On Windows
   winget install uv
   

2. Create and activate a virtual environment:

   uv venv .venv
   source .venv/bin/activate  # On Unix/macOS
   # or
   .venv\Scripts\activate     # On Windows
   

3. Install the required dependencies:

   uv pip install -r pyproject.toml
   

4. Run the server using mcp dev:

   mcp dev main.py
   

5. The server will start and listen on:

   • Host: localhost (127.0.0.1)
   • Port: 8080
   • Transport: Server-Sent Events (SSE)

6. Test the server using tiny agent client:

   tiny-agents run agent.json
   

   The server implements three endpoints:

   • Tool: get_weather(location: str)
   • Resource: weather://{location}
   • Prompt: weather_report(location: str)

7. To stop the server, press Ctrl+C in the terminal

Note

This is a basic example that returns static data. In a real application, you would:

• Connect to actual data sources
• Implement proper error handling
• Add authentication and authorization
• Implement proper logging
• Add comprehensive documentation
• Configure different transport methods beyond SSE
• Add proper server configuration options
• Implement production-ready error handling
• Add monitoring and metrics
• Configure proper port management
• Add environment variable support
• Handle edge cases and validation

This example is intentionally kept simple to focus on MCP concepts rather than production considerations.